Values in (climate) science: What model-based assessments of climate sensitivity teach us about value-judgements and demands on norms thereon

The need for norms ensuring ethical decision-making in policy is well established, extending to decisions made in the scientific practice that informs policy. Values, including non-epistemic ones such as social values, may guide decision-making in the scientific research process where evidence supports more than one decision given uncertainty, and are thus targeted by many normative suggestions from the philosophical literature. How value-judgements enter the body of research that underlies climate change information, with its immediate relevance for urgent mitigation and adaptation decisions, and how the norms may apply here, is however unclear.

In a practical contribution to the debate on values in climate science, we discuss the process of assessing equilibrium climate sensitivity (ECS), an idealised property of the real world of high scientific and societal relevance that has as the ‘holy grail’ of climate science been regularly assessed by the Intergovernmental Panel of Climate Change. We develop a framework consisting of the steps ​model building​, ​deriving ECS​, ​combining model results​, and communicating the findings​ along with the overarching choice of research question​ and ​publishing​, and present and summarise uncertainties, choices, and possible value-judgements involved in each step. We discuss this in the context of scientific objectivity, scrutinise existing normative, action-guiding literature on values, and suggest requirements for applicable norms and ideas.

We find that both epistemic and non-epistemic values are likely to come into play in scientific practice, with the latter arguably playing a relatively larger role further along the assessment steps. A review of existing literature shows that many of the norms proposed do not reflect the characteristics and complexities of assessments drawing on climate modelling: We find that, among others, it is particularly the distribution of epistemic agency; the technical nature of many of the choices; the unpredictability of a decision for further/future model outcomes; the multi-purposeness of models; and the type of value-judgements -other than risk preferences- involved that pose challenges for existing normative ideas. This calls for the development of new such framings more easily applicable to climate science, potentially guided by the insights presented including the step-framework suggested as a way to structure the analysis of the assessment process.

2.11.0.0